| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 1 | <!DOCTYPE html> |
| 2 | <html> |
| 3 | <head> |
| 4 | <title>DeReKo-Word-Vector-Distances: <%= $word %></title> |
| 5 | <link rel="stylesheet" href="//code.jquery.com/ui/1.11.4/themes/smoothness/jquery-ui.css"> |
| 6 | <script src="http://code.jquery.com/jquery-latest.min.js"></script> |
| 7 | <script src="//code.jquery.com/ui/1.11.4/jquery-ui.js"></script> |
| 8 | <script> |
| 9 | $(function() { |
| 10 | $( document ).tooltip({ |
| 11 | content: function() { |
| 12 | return $(this).attr('title'); |
| 13 | }} |
| 14 | ) |
| 15 | }) |
| 16 | </script> |
| 17 | <script src="//d3js.org/d3.v3.min.js" charset="utf-8"></script> |
| Marc Kupietz | adaa163 | 2017-07-04 14:10:29 +0200 | [diff] [blame] | 18 | <script src="/word2vec/js/tsne.js"></script> |
| 19 | <script src="/word2vec/js/som.js"></script> |
| 20 | <script src="/word2vec/js/labeler.js"></script> |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 21 | <style> |
| 22 | body, input { |
| 23 | font-family: Arial, sans-serif; |
| 24 | font-size: 11pt; |
| 25 | } |
| 26 | |
| 27 | .ui-tooltip-content { |
| 28 | font-size: 9pt; |
| 29 | color: #222222; |
| 30 | } |
| 31 | |
| 32 | svg > .ui-tooltip-content { |
| 33 | font-size: 8pt; |
| 34 | color: #222222; |
| 35 | } |
| 36 | |
| 37 | a.merged { |
| 38 | color: green; |
| 39 | fill: green; |
| 40 | } |
| 41 | |
| 42 | #first a { |
| 43 | text-decoration: none; |
| 44 | } |
| 45 | |
| 46 | a.marked, #first a.marked { |
| 47 | text-decoration: underline; |
| 48 | } |
| Marc Kupietz | f4b4939 | 2016-04-28 10:49:56 +0200 | [diff] [blame] | 49 | |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 50 | a.target { |
| 51 | color: red; |
| 52 | fill: red; |
| 53 | } |
| 54 | |
| 55 | #collocators { |
| 56 | margin-bottom: 15px; |
| 57 | } |
| 58 | |
| 59 | #wrapper { |
| 60 | width: 100%; |
| 61 | // border: 1px solid red; |
| 62 | overflow: hidden; /* will contain if #first is longer than #second */ |
| 63 | } |
| 64 | #first { |
| 65 | margin-right: 20px; |
| 66 | float: left; |
| 67 | // border: 1px solid green; |
| 68 | } |
| 69 | #second { |
| 70 | border: 1px solid #333; |
| 71 | overflow: hidden; /* if you don't want #second to wrap below #first */ |
| 72 | } |
| 73 | #som2 svg { |
| 74 | border: 1px solid #333; |
| 75 | } |
| 76 | |
| 77 | #cost { |
| 78 | font-size: 8pt; |
| 79 | color: #222222; |
| 80 | margin-top: 4px; |
| 81 | margin-bottom: 12px; |
| 82 | } |
| 83 | |
| 84 | #sominfo1, #sominfo { |
| 85 | font-size: 8pt; |
| 86 | color: #222222; |
| 87 | margin-top: 0px; |
| 88 | } |
| 89 | |
| 90 | #somcolor1, #somcolor2, #somcolor3 { |
| 91 | display: inline-block; |
| 92 | height: 10px; |
| 93 | width: 10px; |
| 94 | } |
| 95 | |
| 96 | #third { |
| 97 | border: 1px solid #333; |
| 98 | } |
| 99 | |
| 100 | </style> |
| 101 | <script> |
| 102 | |
| 103 | var opt = {epsilon: <%= $epsilon %>, perplexity: <%= $perplexity %>}, |
| 104 | mapWidth = 800, // width map |
| 105 | mapHeight = 800, |
| 106 | jitterRadius = 7; |
| 107 | |
| 108 | var T = new tsnejs.tSNE(opt); // create a tSNE instance |
| 109 | |
| 110 | var Y; |
| 111 | |
| 112 | var data; |
| 113 | var labeler; |
| 114 | |
| 115 | function applyJitter() { |
| 116 | svg.selectAll('.tsnet') |
| 117 | .data(labels) |
| 118 | .transition() |
| 119 | .duration(50) |
| 120 | .attr("transform", function(d, i) { |
| 121 | T.Y[i][0] = (d.x - mapWidth/2 - tx)/ss/20; |
| 122 | T.Y[i][1] = (d.y - mapHeight/2 - ty)/ss/20; |
| 123 | return "translate(" + |
| 124 | (d.x) + "," + |
| 125 | (d.y) + ")"; |
| 126 | }); |
| 127 | } |
| 128 | |
| 129 | function updateEmbedding() { |
| 130 | var Y = T.getSolution(); |
| 131 | svg.selectAll('.tsnet') |
| 132 | .data(data.words) |
| 133 | .attr("transform", function(d, i) { |
| 134 | return "translate(" + |
| 135 | ((Y[i][0]*20*ss + tx) + mapWidth/2) + "," + |
| 136 | ((Y[i][1]*20*ss + ty) + mapHeight/2) + ")"; }); |
| 137 | } |
| 138 | |
| 139 | var svg; |
| 140 | var labels = []; |
| 141 | var anchor_array = []; |
| 142 | var text; |
| 143 | |
| 144 | function drawEmbedding() { |
| 145 | $("#embed").empty(); |
| 146 | var div = d3.select("#embed"); |
| 147 | |
| 148 | // get min and max in each column of Y |
| 149 | var Y = T.Y; |
| 150 | |
| 151 | svg = div.append("svg") // svg is global |
| 152 | .attr("width", mapWidth) |
| 153 | .attr("height", mapHeight); |
| 154 | |
| 155 | var g = svg.selectAll(".b") |
| 156 | .data(data.words) |
| 157 | .enter().append("g") |
| 158 | .attr("class", "tsnet"); |
| 159 | |
| 160 | g.append("a") |
| Marc Kupietz | f4b4939 | 2016-04-28 10:49:56 +0200 | [diff] [blame] | 161 | .attr("xlink:href", function(word) { |
| 162 | return (data.urlprefix+word);}) |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 163 | .attr("class", function(d, i) { |
| 164 | var res=""; |
| 165 | if(data.marked[i]) { |
| 166 | res="marked "; |
| 167 | } |
| 168 | if(data.target.indexOf(" "+d+" ") >= 0) { |
| 169 | return res+"target"; |
| 170 | } else if(data.ranks[i] < data.mergedEnd) { |
| 171 | return res+"merged"; |
| 172 | } else { |
| 173 | return res; |
| 174 | } |
| 175 | }) |
| 176 | .attr("title", function(d, i) { |
| 177 | if(data.mergedEnd > 0) { |
| 178 | if(data.ranks[i] >= data.mergedEnd) { |
| 179 | return "rank: "+i +" "+"freq. rank: "+(data.ranks[i]).toString().replace(/\B(?=(\d{3})+(?!\d))/g, ","); |
| 180 | } else { |
| 181 | return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ",") + " (merged vocab)"; |
| 182 | } |
| 183 | } else { |
| 184 | return "rank: "+i +" "+"freq. rank: "+data.ranks[i].toString().replace(/\B(?=(\d{3})+(?!\d))/g, ","); |
| 185 | } |
| 186 | }) |
| 187 | .append("text") |
| 188 | .attr("text-anchor", "top") |
| 189 | .attr("font-size", 12) |
| 190 | .text(function(d) { return d; }); |
| 191 | |
| 192 | var zoomListener = d3.behavior.zoom() |
| 193 | .scaleExtent([0.1, 10]) |
| 194 | .center([0,0]) |
| 195 | .on("zoom", zoomHandler); |
| 196 | zoomListener(svg); |
| 197 | } |
| 198 | |
| 199 | var tx=0, ty=0; |
| 200 | var ss=1; |
| 201 | var iter_id=-1; |
| 202 | |
| 203 | function zoomHandler() { |
| 204 | tx = d3.event.translate[0]; |
| 205 | ty = d3.event.translate[1]; |
| 206 | ss = d3.event.scale; |
| 207 | updateEmbedding(); |
| 208 | } |
| 209 | |
| 210 | var stepnum = 0; |
| 211 | |
| 212 | function stopStep() { |
| 213 | clearInterval(iter_id); |
| 214 | text = svg.selectAll("text"); |
| 215 | |
| 216 | // jitter function needs different data and co-ordinate representation |
| 217 | labels = d3.range(data.words.length).map(function(i) { |
| 218 | var x = (T.Y[i][0]*20*ss + tx) + mapWidth/2; |
| 219 | var y = (T.Y[i][1]*20*ss + ty) + mapHeight/2; |
| 220 | anchor_array.push({x: x, y: y, r: jitterRadius}); |
| 221 | return { |
| 222 | x: x, |
| 223 | y: y, |
| 224 | name: data.words[i] |
| 225 | }; |
| 226 | }); |
| 227 | |
| 228 | // get the actual label bounding boxes for the jitter function |
| 229 | var index = 0; |
| 230 | text.each(function() { |
| 231 | labels[index].width = this.getBBox().width; |
| 232 | labels[index].height = this.getBBox().height; |
| 233 | index += 1; |
| 234 | }); |
| 235 | |
| 236 | |
| 237 | // setTimeout(updateEmbedding, 1); |
| 238 | // setTimeout( |
| 239 | labeler = d3.labeler() |
| 240 | .label(labels) |
| 241 | .anchor(anchor_array) |
| 242 | .width(mapWidth) |
| 243 | .height(mapHeight) |
| 244 | .update(applyJitter); |
| 245 | // .start(1000); |
| 246 | |
| 247 | iter_id = setInterval(jitterStep, 1); |
| 248 | } |
| 249 | |
| 250 | var jitter_i=0; |
| 251 | |
| 252 | function jitterStep() { |
| 253 | if(jitter_i++ > 100) { |
| 254 | clearInterval(iter_id); |
| 255 | } else { |
| 256 | labeler.start2(10); |
| 257 | applyJitter(); |
| 258 | } |
| 259 | } |
| 260 | |
| 261 | var last_cost=1000; |
| 262 | |
| 263 | function step() { |
| 264 | var i = T.iter; |
| 265 | |
| 266 | if(i > <%= $no_iterations %>) { |
| 267 | stopStep(); |
| 268 | } else { |
| 269 | var cost = Math.round(T.step() * 100000) / 100000; // do a few steps |
| 270 | $("#cost").html("tsne iteration " + i + ", cost: " + cost.toFixed(5)); |
| 271 | if(i % 250 == 0 && cost >= last_cost) { |
| 272 | stopStep(); |
| 273 | } else { |
| 274 | last_cost = cost; |
| 275 | updateEmbedding(); |
| 276 | } |
| 277 | } |
| 278 | } |
| 279 | |
| 280 | function showMap(j) { |
| 281 | data=j; |
| 282 | T.iter=0; |
| 283 | T.initDataRaw(data.vecs); // init embedding |
| 284 | drawEmbedding(); // draw initial embedding |
| 285 | |
| 286 | if(iter_id >= 0) { |
| 287 | clearInterval(iter_id); |
| 288 | } |
| 289 | //T.debugGrad(); |
| 290 | iter_id = setInterval(step, 1); |
| 291 | if(<%= $show_som %>) { |
| 292 | makeSOM(j, <%= $no_iterations %>); |
| 293 | } |
| 294 | } |
| 295 | |
| 296 | </script> |
| 297 | </head> |
| 298 | <body> |
| Marc Kupietz | b3422c1 | 2017-07-04 14:12:11 +0200 | [diff] [blame^] | 299 | <form method="GET"> |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 300 | word(s): |
| 301 | <input type="text" name="word" size="20" value="<%= $word %>" title="When looking for multiple words use spaces as separators to search around the average vector and | as separator to get the neighbours for each word."> |
| 302 | % if($mergedEnd > 0) { |
| 303 | backw. <input type="checkbox" name="sbf" value="1" <%= ($searchBaseVocabFirst ? "checked" : "") %> title="If checkecked base vocabulary will be searched first. Otherwise merged vocabulray will be searched first."> |
| 304 | % } |
| 305 | max. neighbours: <input type="text" size="8" name="n" value="<%= $no_nbs %>"> |
| 306 | max. iterations: <input type="text" name="N" size="8" value="<%= $no_iterations %>"> |
| 307 | SOM <input type="checkbox" name="som" value="1" <%= ($show_som ? "checked" : "") %>> |
| 308 | % if($collocators) { |
| 309 | <span> </span>sort collocators by |
| 310 | <select name="sort"> |
| 311 | <option value="0" <%= ($sort!=1? "selected":"") %>>responsiveness</option> |
| 312 | <option value="1" <%= ($sort==1? "selected":"") %>>mean p</option> |
| 313 | </select> |
| 314 | % } |
| 315 | <span> </span><input type="submit" value="Show"> |
| 316 | </form> |
| 317 | <br> |
| 318 | % if($lists) { |
| 319 | <div id="wrapper"> |
| 320 | <table id="first"> |
| 321 | <tr> |
| 322 | <th align="right">#</th><th align="right">cos</th><th align="left">paradigmatic</th> |
| 323 | % if($collocators) { |
| 324 | <th title="Position in winodw around target word. Absolute value can be too low because of sub-sampling frequent words.">@</th><th align="right" title=""Responsivenes" of the collocator at the relative position @. Approximation of the probability that the combination of the target word and the collocator at the relative position @ come from the corpus.">resp.</th><th title="Probability of the collocator at window location @."align="right">p(c<sub><small>@</small></sub>)</th><th align="right">Σp(c<sub><small>@</small></sub>)/|w|</th><th align="left">syntagmatic</th> |
| 325 | % } |
| 326 | </tr> |
| 327 | % my $j=0; my @words; my @vecs; my @ranks; my @marked; |
| 328 | % for my $list (@$lists) { |
| 329 | % my $i=0; while($list) { |
| 330 | % my $item = (@$list)[$i]; |
| 331 | % my $c = ($collocators? (@$collocators)[$i] : 0); |
| 332 | % last if(!$c && !$item); |
| 333 | <tr> |
| 334 | <td align="right"> |
| 335 | <%= ++$i %>. |
| 336 | </td> |
| 337 | % if($item) { |
| 338 | % if(!grep{$_ eq $item->{word}} @words) { |
| 339 | % push @vecs, $item->{vector}; |
| 340 | % push @words, $item->{word}; |
| 341 | % push @ranks, $item->{rank}; |
| 342 | % push @marked, ($marked->{$item->{word}}? 1 : 0); |
| 343 | % } |
| 344 | <td align="right"> |
| 345 | <%= sprintf("%.3f", $item->{dist}) %> |
| 346 | </td> |
| 347 | <td> |
| 348 | % my $class = ($marked->{$item->{word}}? "marked " : ""); |
| 349 | % my $r = $item->{rank}; |
| 350 | % if($r < $mergedEnd) { |
| 351 | % $class .= "merged"; |
| 352 | % $r .= " (merged vocab)"; |
| 353 | % } elsif($mergedEnd!=0 && $r > $mergedEnd) { |
| 354 | % $r -= $mergedEnd; |
| 355 | % } |
| Marc Kupietz | f4b4939 | 2016-04-28 10:49:56 +0200 | [diff] [blame] | 356 | <a class="<%= $class =%>" |
| 357 | title="freq. rank: <%= $r =%>" |
| 358 | href="<%= url_with->query([word => $item->{word}]) =%>"> |
| 359 | <%= $item->{word} =%> |
| 360 | </a> |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 361 | </td> |
| 362 | % } else { |
| 363 | <td colspan="2"/> |
| 364 | % } |
| 365 | % if($c) { |
| 366 | <td align="right"> |
| 367 | <%= $c->{pos} %>: |
| 368 | </td> |
| 369 | <td align="right"> |
| 370 | <%= sprintf("%.3f", $c->{dist}) %> |
| 371 | </td> |
| 372 | <td align="right"> |
| 373 | <%= sprintf("%.3e", $c->{norm}) %> |
| 374 | </td> |
| 375 | <td align="right"> |
| 376 | <%= sprintf("%.3e", $c->{sum}) %> |
| 377 | </td> |
| 378 | <td align="left"> |
| Marc Kupietz | f4b4939 | 2016-04-28 10:49:56 +0200 | [diff] [blame] | 379 | <a href="<%= url_with->query([word => $c->{word}]) =%>"> |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 380 | <%= $c->{word} %> |
| 381 | </td> |
| 382 | % } else { |
| 383 | <td colspan="5"/> |
| 384 | % } |
| 385 | </tr> |
| 386 | % } |
| 387 | % } |
| 388 | </table> |
| 389 | <script> |
| 390 | % use Mojo::ByteStream 'b'; |
| Marc Kupietz | f4b4939 | 2016-04-28 10:49:56 +0200 | [diff] [blame] | 391 | % my $urlprefix = url_with->query([word=>'']); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 392 | $(window).load(function() { |
| Marc Kupietz | f4b4939 | 2016-04-28 10:49:56 +0200 | [diff] [blame] | 393 | showMap(<%= b(Mojo::JSON::to_json({target => " $word ", mergedEnd=> $mergedEnd, words => \@words, vecs => \@vecs, ranks => \@ranks, marked => \@marked, urlprefix => $urlprefix})); %>); |
| Marc Kupietz | 8330522 | 2016-04-28 09:57:22 +0200 | [diff] [blame] | 394 | }); |
| 395 | </script> |
| 396 | % } |
| 397 | <div id="second" style="width:800px; height:800px; font-family: arial;"> |
| 398 | <div id="embed"> |
| 399 | </div> |
| 400 | </div> |
| 401 | <div id="cost"></div> |
| 402 | % if($show_som) { |
| 403 | <div id="som2"> |
| 404 | </div> |
| 405 | <div id="sominfo1"><span id="somcolor1"> </span> <span id="somword1"> </span> <span id="somcolor2"> </span> <span id="somword2"> </span> <span id="somcolor3"> </span></div> |
| 406 | <div id="sominfo">SOM iteration <span id="iterations">0</span></div> |
| 407 | % } |
| 408 | </div> |
| 409 | % if($training_args) { |
| 410 | <p> |
| 411 | Word vector model trained with <a href="https://code.google.com/p/word2vec/">word2vec</a> using the following parameters: <pre><%= $training_args %></pre> |
| 412 | </p> |
| 413 | % } |
| 414 | </body> |
| 415 | </html> |